Designing personalized incentive-based demand response services based on smart meter data and NSGA-III-DE algorithm

• Published in: Energy (Oxford) Vol. 334; p. 137454
• Main Authors: Qin, Chenxi, Liu, Wenjie, Yan, Yuejun, Guan, Quanxue, Wang, Qin
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 15.10.2025
• Subjects: Customer behavior analysis; Demand response; NSGA-III-DE; Personalized service; Smart meter data; Customer behavior analysis; Demand response; Smart meter data; Personalized service; NSGA-III-DE
• ISSN: 0360-5442
• DOI: 10.1016/j.energy.2025.137454

During peak demand or urgency periods, power systems may face challenges due to insufficient electricity supply. One practical approach to addressing this issue is incentive-based demand response (IBDR). In this approach, residential customers participate in the IBDR programs by preemptively signing contracts with load aggregators (LAs) and adjusting their electricity consumption during peak periods in exchange for incentive subsidies. This paper proposes a method for designing personalized IBDR services by analyzing electricity consumption data and solving a multi-objective optimization problem. We analyze smart meter data using an adaptive K-means clustering algorithm combined with a fuzzy system to understand customers’ electricity consumption preferences. Additionally, we employ a stacked biGRU-biLSTM model with an attention mechanism for load forecasting to understand electricity usage during responsive periods. Subsequently, we introduce a multi-objective optimization model aimed at maximizing the response quantity while simultaneously mitigating customer discomfort and reducing the operational costs of LAs. Following this, the NSGA-III-DE algorithm is employed to design personalized IBDR services for enhanced participation and implementation effectiveness. In the numerical simulations, we observe that by offering personalized IBDR services, LA’s electricity procurement expenditures were successfully reduced by 50%. Moreover, there was a significant increase in residential customers’ enthusiasm to participate in the demand response program, with a response rate reaching 85% of the total potential. These results clearly demonstrate the effectiveness of the proposed method. •A novel IBDR scheme that addresses inconsistent customer responses is proposed.•A novel deep learning model and adaptive K-means clustering extract usage patterns.•Dissatisfaction functions tailored to observed consumption behaviors are proposed.•A multi-objective optimization model to design personalized IBDR services is created.•The efficiency of personalized IBDR services is analyzed using real-world data.